The bloodiest and most significant battle of the American Civil War took place in Pennsylvania. At the outset of that conflict, the forces of the North - greater in number and better armed - were regarded as the overwhelming favourites to win the struggle. Yet they were outsmarted by the charismatic General Robert E. Lee, who proved to be more imaginative in the field, inspiring passionate loyalty in his Confederate soldiers.

A bemused Abraham Lincoln was reduced to hiring and firing his generals and constantly reshaping his strategy. By July 1863, it seemed as if the Confederates might storm Washington itself and pull off an extraordinary victory. Their momentum was, however, halted by three days of combat on the fields of Gettysburg. Lee was forced to retreat and his reputation for invincibility was ended. From there, the machine that was the Union slowly but surely crushed its opponents.

Hillary Clinton must hope that history repeats itself in Pennsylvania tomorrow. Having started as the clear frontrunner for the Democratic nomination, she has been stunned by Barack Obama's ability to portray himself as the agent of political change, his skill at motivating activists and success in employing the internet to break fundraising records.

He has been the General Lee of the competition so far. If he were to win the Pennsylvania primary, he would indeed become unstoppable. Yet adversity has brought out the best in Mrs Clinton. She has fought for seven weeks in Pennsylvania and while no one has been killed or wounded (unlike the 8,000 dead and near 50,000 casualties and losses at Gettysburg) it has been a bruising struggle with Mrs Clinton landing the most blows. The odds are that she will at least emerge strong enough to take her cause on further.

Whether she can emulate Lincoln or not, though, depends on Democratic “superdelegates”. Neither she nor Mr Obama can secure an overall majority out of those pledged delegates who have been selected in the various primaries and caucuses: there are not enough delegates in the nine skirmishes left after Pennsylvania for that to happen. Everything will thus turn on the almost 800 individuals who have a vote by dint of their present or past service as Presidents or Vice-Presidents, in Congress, as governors of states, as members of the Democratic National Committee or other form of local worthies such as mayors. Strictly speaking, these people are known either as “party leaders and elected officials” (PLEOs) or “un-pledged add-on delegates” (UPADs). You can see why they prefer being called “superdelegates”.

The chances are that Mr Obama will end the nomination season with more pledged delegates than Mrs Clinton. His admirers argue that it would be profoundly wrong for those who have not been elected as delegates to overturn the will of those who have. It's a seductive claim, but there are good reasons why the superdelegates should ignore it and instead endorse Mrs Clinton.

The first is, what is the point of the superdelegate system if all they do is follow the majority of pledged delegates? Why bother with them? Why not just allow them to turn up at the convention as mere observers? The Democratic Party created the superdelegate system about 25 years ago because it feared that the party's most ideological supporters were quite capable of choosing a candidate who many ordinary Democrats would not feel able to back at polling stations. If the primaries and caucuses were to be the gearbox of the nominating procedure, then the superdelegates were designed to serve as the handbrake. That is their role.

Secondly, any advantage that Mr Obama will have among pledged delegates is misleading. Not only will Mrs Clinton have won in most of the largest states but she will probably have secured the bulk of delegates won in primaries - where turnout is comparatively high, while he has romped home in the caucuses - where participation is notoriously feeble.

Furthermore, if all the superdelegates were compelled to vote for the person who won the most votes in their state (which they should not be, but it is an interesting exercise), then Mrs Clinton, who is likely to end the season having triumphed in eight of the most populous ten states (including Florida and Michigan, which had their results discounted by the Democratic National Committee as punishment for scheduling their primaries too early), would benefit hugely.

Finally, enough is now known about the strengths and weaknesses of these two contenders for superdelegates to come to the following conclusion. Mrs Clinton is the 5347 option and Mr Obama is the 5542 one. By this I mean that it is tough to imagine her obtaining more than 53 per cent of the national vote against John McCain, but it is hard to envisage her falling below 47 per cent either. Most of those Democrats who prefer Mr Obama to her (African-Americans, affluent whites, the young) would nevertheless back the New York senator in November (particularly if their man was in the vice-presidential slot).

Mr Obama, by contrast, has a somewhat higher vote ceiling but a much lower floor to his vote. If Americans decide that they are desperate for “change”, pure and simple, then he is a better vehicle for that mood than a woman who has the history of the 1990s attached to her. If, though, voters are after “change (with reassurance)”, as one suspects is the case, then she is a smarter bet against Mr McCain. A sizeable slice of working-class Democrats who back her may switch to the Arizona Senator if she loses. In the worst-case scenario, the Republican champion may well wipe the floor with Mr Obama.

Assuming she is victorious in Pennsylvania, then Mrs Clinton should keep on running. The superdelegates must ask themselves not only “who can win?” but “how might they lose?” For the reality of Gettysburg is not that in pure military terms the North actually won, but that it did not lose. It was this that later made it such a decisive moment.

Times Recommends

• We have kit and leadership. Let’s go
• Behind the scenes at the biggest deal
• Conferences — what a waste of money